Learning and Individual Differences 77 (2020) 101811

Contents lists available at ScienceDirect

Learning and Individual Differences

journal homepage: www.elsevier.com/locate/lindif

Academic achievement and depressive symptoms: Are fixed mindsets T

distinct from negative attributional style?
⁎
Andrea I. Alatorre, Rosalie V. DePaola, Gerald J. Haeffel
Department of Psychology, University of Notre Dame, United States of America

A R T I C LE I N FO A B S T R A C T

Keywords: Over the last thirty years, there has been an explosion of research on “mindset” theories of personal attributes
Attributional style such as intelligence (Dweck, 1988). Research shows that individuals who believe that their attributes are fixed
Hopelessness theory (rather than changeable) are more likely to have negative academic outcomes and negative moods. However, it
Entity remains unclear if these mindset theories represent a novel construct or if they overlap with the cognitive
Incremental
theories of depression which were theorized a decade earlier. According to the cognitive theories, people who
Mindsets
attribute negative life events to stable and global (i.e., unchangeable) causes are more likely to become helpless
and develop depression. The purpose of the current study was to test whether the newer mindset theories
provide an incremental advance in knowledge or are more likely specific instances of a more general negative
attributional style. To this end, we conducted a two time-point prospective study with 130 undergraduates.
Contrary to hypotheses, results showed that intelligence mindset was distinct from negative attributional style.
However, intelligence mindset did not demonstrate predictive or incremental validity; it did not predict end of
semester GPA, dropping a class or depressive symptoms. In contrast, negative attributional style predicted both
dropping a class and increases in depressive symptoms. Theoretical and practical implications of these findings
are discussed.

According to “mindset” theories (also called implicit theories), may view the same setbacks as opportunities to develop new skills and
people differ in their perceptions of the malleability of personal attri- build intelligence.
butes such as personality and intelligence. These individual differences Research has generally supported the mindset theory of intelligence.
in beliefs about malleability (also known as mindsets) are hypothesized According to a meta-analysis by Burnette, O'Boyle, VanEpps, Pollack,
to be related to academic and emotional outcomes. According to Dweck and Finkel (2013), “…mind-sets matter. That is, implicit theories are
(1999), individual's mindsets form a framework for interpreting and indeed consequential for self-regulatory processes and goal achieve-
responding to adversity (Dweck, 1999; Dweck & Leggett, 1988; Molden ment” (p. 680). Prospective longitudinal studies consistently find that
& Dweck, 2006). One specific mindset that has received a considerable those with greater entity than incremental beliefs about intelligence
amount of empirical attention is that of intelligence (see Sisk, have worse academic outcomes; the effect sizes found in these studies
Burgoyne, Sun, Butler, & Macnamara, 2018 for review). The in- range from small to medium. For example, Blackwell, Trzesniewski, and
telligence mindset theory has two components that lie on a continuum: Dweck (2007) found that 7th graders with greater entity mindset of
entity and incremental. Individuals with an entity mindset believe that intelligence had a flat trajectory of academic achievement (grades)
intelligence is a fixed trait and difficult to change, whereas those with throughout junior high school whereas as those with greater incre-
an incremental mindset believe that intelligence is malleable and can be mental mindset had an upward trend in grades throughout that same
changed with effort. Having greater entity than incremental beliefs time period (see Henderson & Dweck, 1990 for a similar finding). In
about intelligence is thought to be maladaptive because it encourages addition to predicting concrete quantitative outcomes like grades and
people to believe that they cannot change their intelligence, thus lim- GPA, individuals with greater incremental mindset are more likely than
iting a person's ability for intellectual growth and improvement. For those with greater entity mindset to focus on goals aimed at increasing
example, students with greater entity beliefs of intelligence may attri- their ability as opposed to goals aimed at documenting their ability
bute academic setbacks and challenges to a low level of intelligence (Dweck & Leggett, 1988). This indicates that a person's mindset not
that is hopeless to change, but those with greater incremental beliefs only influences quantitative outcomes but also qualitative outcomes

⁎
Corresponding author at: 390 Corbett Hall, University of Notre Dame, Notre Dame, IN 46656, United States of America.
E-mail address: ghaeff[email protected] (G.J. Haeffel).

https://doi.org/10.1016/j.lindif.2019.101811
Received 8 March 2019; Received in revised form 3 December 2019; Accepted 10 December 2019
1041-6080/ © 2019 Elsevier Inc. All rights reserved.
A.I. Alatorre, et al. Learning and Individual Differences 77 (2020) 101811

that impact the way in which they think about their ability to change similarly negative inferences in response to a variety of other life si-
their intelligence. It is important to note, however, that studies using tuations. Our hypothesis is also justified by the fact that mindset the-
experimental designs (in which mindsets are directly manipulated) tend ories were derived from Dweck's (1986) (e.g., Tsujimoto et al., 2018;
to show smaller effect sizes and more circumscribed effects than the Weiner, 1985; Wilson & Linville, 1982) seminal work on motivational
longitudinal correlational studies in this area (Yeager et al., 2018). In processes in learning, which were based partially on the attribution
intervention studies, the positive findings tend to hold in low, but not theories of depression (specifically, learned helplessness theory). This
high, achieving students. early work focused on understanding children's attributions for failure
Over the last decade, research on mindsets has surged in popularity (i.e., attributing poor academic performance to effort or ability). Fur-
and has also grown in scope. Recent mindset work not only focuses on ther, the interventions created by Dweck borrowed directly from cog-
beliefs about intelligence, but also a variety of other human char- nitive therapy for depression (also based on the cognitive theories of
acteristics such as personality and morality. It also been extended to depression), which focuses on the cognitive restructuring of attribu-
include predictions regarding emotional outcomes. Entity beliefs about tions. It was only more recently that the emphasis changed to children's
personal characteristics are thought to be related to negative emotional beliefs/mindsets about the stability of self-concepts such as intelligence
outcomes such as depression and anxiety (Burnette et al., 2013). For and personality.
example, a recent meta-analysis showed a positive association between It remains unclear if the proliferation of these mindset theories are
entity theories of personal traits—namely intelligence, personality, and novel contributions to the field or a reinvention of work derived from
peer relationships—and youth mental health problems (Schleider, Abel, attributional theory. For example, Schroder, Moran, Donnellan, and
& Weisz, 2015). Similarly, Miu and Yeager (2014) showed that a brief Moser (2016) found that although the different mindset domains (e.g.,
intervention teaching an incremental mindset of personality reduced intelligence, personality, morality, etc.) are distinguishable from one
the incidence of clinically significant levels of self-reported depressive another, it appears that there is a “global dimension that captures
symptoms nine months post-intervention. whether a person tends to adopt a growth versus fixed mind-set re-
Despite the popularity and increasing number of studies on mind- gardless of domain.” We contend that this more global dimension is a
sets, it remains unclear if the mindset hypothesis represents a novel negative attributional style. This may explain why prospective long-
advancement for understanding academic and affective responses or if itudinal designs examining the correlation between mindset and aca-
it overlaps with previous work on cognition and emotion. In the 1970s, demic outcomes yield more consistent and stronger effects than ex-
researchers proposed the cognitive theories of depression (e.g., periments in which mindset is manipulated. Interventions targeting a
Abramson, Seligman, & Teasdale, 1978; Beck, 1976; Ellis, 1962). Ac- particular mindset may not be effective if that mindset represents a
cording to the hopelessness theory of depression (Abramson, Metalsky, more general underlying style (that is not modified by the interven-
& Alloy, 1989), people who attribute negative life events to stable tion).
(consistent over time) and global (affects many areas of their life) In summation, we contend that intelligence mindset is a subset of a
causes are more likely to develop hopelessness and depressive symp- negative attributional style. We hypothesize that people who tend to
toms than people who generate unstable, specific attributions about believe their intelligence is stable have a more general tendency to
negative life events. In other words, people who attribute the causes of attribute other self-concepts and the causes of events in their lives as
negative events to fixed and unchangeable factors are at heightened risk stable and unchangeable. Thus, we predicted that intelligence mindset
for depression. For example, if a student attributes a bad test grade to and negative attributional style would be highly correlated. Second, we
low intelligence, and they believe that intelligence is stable over time predicted that both intelligence mindset and negative attributional style
(i.e., unchangeable) and global, then they are at heightened risk for would predict changes in end of semester grade point average (GPA),
depression (see Haeffel et al., 2008 for review). This negative attribu- whether or not a student dropped a course, and changes in depressive
tional style vulnerability factor appears to overlap extensively with symptoms during high stress (King, 2017). However, intelligence
mindset theories. Indeed, an examination of the items on the in- mindset would not exhibit incremental predictive validity. That is, we
telligence mindset questionnaire shows that participants are specifically did not expect intelligence mindset to predict unique variance (un-
asked to rate the extent to which they believe intelligence is stable/ accounted for by negative attributional style) in academic and mood
unchangeable over time (e.g., “You have a certain amount of in- outcomes when the two constructs were tested simultaneously in a
telligence, and you really can't do much to change it”; “Your in- regression equation.
telligence is something about you that you can't change very much”; We hypothesized that negative attributional style, a risk factor for
“You can learn new things, but you can't really change your basic in- depression, would predict academic outcomes for a couple of reasons.
telligence”). First, research shows that depression is associated with academic im-
The purpose of the current study was to test whether or not the pairment. Students with depressive symptoms report decreased interest
newer mindset theories represent a distinct construct or are more likely in school, more missed classes, and decreased academic productivity
specific instances of a more general negative attributional style (as (e.g., Heiligenstein & Guenther, 1996). Hysenbegasi, Hass, and
theorized a decade earlier by the cognitive theories of depression). We Rowland (2005) found that depression was associated with a half a
theorize that the negative effects of having an entity mindset of in- letter grade decrease in student GPA. Second, there are at least two
telligence are not necessarily due to beliefs about intelligence per se, studies showing that negative attributional style predicts poor academic
but rather because people with an entity mindset also have a more outcomes. Peterson and Barrett (1987) showed that college freshmen
general negative attributional style. In other words, measures of with a negative attributional style were less likely to have specific
mindsets are inadvertently identifying individuals who have a negative academic goals, less likely to seek academic advising, and had lower
attributional style. Even though the mindset measures do not assess all grades than freshmen with a more adaptive attributional style (i.e.,
of the elements featured in the attributional theories of depression, we those who made unstable causal attributions). Similarly, Nolen-
suspect that individuals who make stable attributions about intelligence Hoeksema, Girgus, and Seligman (1986) showed that negative attri-
are also inclined to make similarly stable (and also global) attributions butional style was associated with achievement related problems.
about other factors in their world. Indeed, prior research shows that the Specifically, they report that negative attributional style predicted
different facets of cognitive vulnerability tend to correlate and load children's self-reported helpless behaviors in the classroom as well as
onto a general negative cognitive vulnerability factor (Haeffel, 2010; teachers' ratings of helplessness. These results are consistent with our
Haeffel et al., 2008; Hankin, Lakdawalla, Carter, Abela, & Adams, hypothesis that the explanatory power of intelligence mindset to pre-
2007). This suggests that those who make stable attributions about a dict negative academic outcomes could be due to it being a proxy for
specific area of their life such as intelligence, are also generating negative attributional style.

2
A.I. Alatorre, et al. Learning and Individual Differences 77 (2020) 101811

To test our hypotheses, we conducted a two time-point prospective effect.

study with 130 undergraduate participants. Analyses tested the corre-
lation between intelligence mindset and negative attributional style as 1.2. Measures
well as their incremental validity (when pitted against each other) for
predicting academic outcomes and changes in depressive symptoms. 1.2.1. Negative attributional style
Such comparisons can help constrain the production of constructs and The Cognitive Style Questionnaire (CSQ; Haeffel et al., 2008) was
theories that overlap with existing constructs and theories as well as used to assess negative attributional style (as featured in the hope-
lead to important integrations of parallel literatures. Philosophers of lessness theory of depression). It is the most commonly used measure of
science such as Meehl (1978) (see also Popper, 1959) have argued that the negative attributional style construct featured in the hopelessness
psychology has failed to develop a cumulative knowledge base, in part, theory of depression (Haeffel et al., 2008). The CSQ is a self-report
because of the proliferation of theories. Meehl (1978) has stated that in questionnaire that presents participants with 12 hypothetical negative
psychology, “…theories rise and decline, come and go, more as a events (6 achievement and 6 interpersonal). Participants imagine the
function of baffled boredom than anything else; and the enterprise events happening to themselves and then make ratings on the three
shows a disturbing absence of that cumulative character…” (p. 807). vulnerability dimensions featured in the hopelessness theory of de-
This indifference has created a field in which theories are studied in pression – stability and globality, probable consequences of each event,
isolation, overlap in conceptual and predictive value, and are never and the self-worth implications of each event. An individual's CSQ score
refuted because the same concepts continue to be reinvented over time. is their average rating across these three dimensions (stability and
Pitting theories against one another, like in the current study, is one globality, consequences, and self-worth characteristics) for the 12 hy-
strategy for building knowledge and possibly falsifying redundant pothetical negative life events. The current study focused on the CSQ
theories. achievement subscale as it is most applicable to academic outcomes.
Scores can range from 1 to 7, with higher scores reflecting greater levels
1. Method of negative attributional style. The CSQ has good internal consistency,
reliability, and validity (see Haeffel et al., 2008 for review). Coefficient
1.1. Participants alpha for the CSQ achievement subscale was 0.86.

Participants were 130 undergraduates (94 female, 36 male; M

1.2.2. Depressive symptoms
age = 18.91, SD = 1.08) recruited from a medium sized private uni-
The Beck Depression Inventory (BDI; Beck, Rush, Shaw, & Emery,
versity in the Midwestern United States. All participants were volun-
1979) was used to assess depressive symptoms. The BDI is a 21-item
teers from the University's Psychology research participant pool.
self-report questionnaire. Scores are created by summing the items
Students enrolled in psychology courses are required to enroll in at least
(range 0–63) with higher scores indicating greater levels of depressive
one of the available psychology studies listed online; students received
symptoms. The BDI has demonstrated strong reliability and validity
either course credit or extra credit points for their participation. The
(Beck, Steer, & Garbin, 1988). Coefficient alpha for the BDI was 0.87 at
ethnicity of the sample was: 64% Caucasian, 20% Hispanic, 13% Asian,
Time 1 and 0.88 at Time 2.
3% African American. Nearly half of the participants (n = 53) reported
their major as psychology or neuroscience and behavior. Business was
1.2.3. Implicit theories of intelligence
the next most popular major (n = 17), followed by the pre-professional
Consistent with prior research, the Theory of Intelligence (TOI;
medical school major (n = 15), and “undecided” (n = 8). The average
Hong, Chiu, Dweck, Lin, & Wan, 1999) was used to measure individuals'
ACT score reported in this sample was 33, which is greater than the
propensity to have an entity mindset of intelligence or an incremental
average ACT score of 21 in the United States (ACT, Inc., 2016). The
mindset of intelligence. Participants were asked to consider 6 state-
average GPA in this sample was 3.48 (“B+” average), which is similar
ments about their beliefs about their intelligence and rate how strongly
to the average GPA of 3.30 for students attending private four-year
they agree or disagree on a 6-point scale. There are no cut-offs for in-
colleges in the United States (Rojstaczer, Stuar, & Healy, 2010).
cremental or entity categories. Rather, the scale is analyzed as a con-
All procedures were approved by the institution's human subject
tinuous construct with scores ranging from 0 to 6. Ratings were coded
review board and consistent with human subject guidelines (partici-
and reverse coded so that a higher score indicates more of an incre-
pants were all volunteers and provided informed consent). The ques-
mental mindset (i.e., a belief that intelligence is malleable). Coefficient
tionnaire was completed confidentially, but not anonymously.
alpha for the TOI was 0.94.
Participants included a phone number in the case that they needed to be
contacted due to high levels of suicidal ideation. In accordance with the
IRB protocol, participants who reported high levels of suicidality (a 1.2.4. Academic outcomes
score of “3” on the suicide item of the BDI) were contacted by a clinical Grade point average (GPA) and whether or not a participant
psychologist (in this case, the PI) to ensure that they were not at im- dropped a course during the study time frame were assessed. They re-
minent danger of self-harm, and then given contact information for ported whether or not they dropped a course at Time 2 (the end of the
three mental health facilities (one on-campus and two off-campus). No semester). Participants rated the item as either “yes” or “no” (responses
participants met the criteria for being contacted. After the BDI was were then coded as 0 or 1; 0 = did not drop a course and 1 = did drop
checked for suicidality scores, any identifying information was deleted a course). Overall GPA was self-reported at the start of the study and
making the data set anonymous. end of semester GPA was obtained from the registrar. GPA scores
ranged from 0 to 4 (0 = “F”, 1 = “D”, 2 = “C”, 3 = “B”, 4 = “A”).
1.1.1. Power analysis
The sample size of 130 was used for the statistical power analyses 1.2.5. Academic stress
(Zhang & Yuan, 2015) and a 2-predictor variable equation was used as a Participants self-reported if they were “experiencing significant
baseline (main effect of TOI or CSQ and baseline control variable for the stress about their academic performance this semester.” Participants
T2 dependent measure). The recommended effect sizes used for this rated the item as either “yes” or “no”. Academic stress was assessed in
assessment using multiple regression were as follows: small order to test the negative attributional style by stress interaction hy-
(R2 = 0.05), medium (R2 = 0.15), and large (R2 = 0.26). The alpha pothesis. The hopelessness theory posits a vulnerability-stress interac-
level used for this analysis was p < .05. The analysis showed that the tion by which negative attributional style is most likely to lead to de-
statistical power for this study was 0.80 for detecting a small to medium pressive symptoms in the presence but not absence of stress.

3
A.I. Alatorre, et al. Learning and Individual Differences 77 (2020) 101811

1.3. Procedure 2.2. Predicting end of semester GPA

The study used a 3-month prospective longitudinal design. The We hypothesized that individuals with greater entity beliefs (i.e.,
three-month time frame was used for two primary reasons. First, this lower TOI scores) and greater levels of negative attributional style (i.e.,
time frame mapped onto the length of the academic semester. It was higher CSQ achievement subscale scores) would be more likely to have
necessary to administer measures at the start of classes, and then again lower end of semester GPAs than those with greater incremental beliefs
at the end of the semester after grades were received. Second, the time of intelligence and less negative attributional style. Hierarchical mul-
frame provided ample time for individual differences to emerge in the tiple regression was used to test the hypothesis (see Table 2). Results
outcome variables of interest (specifically, decisions to drop a course as showed that neither the TOI nor CSQ achievement subscale were sig-
well as changes in depressive symptoms). At the University in which the nificant predictors of end of semester GPA. The only significant pre-
study was conducted, students are allowed to drop a course anytime dictor of end of semester GPA was baseline GPA. As expected, higher
during the first two months of the semester. The three-month time overall GPA scores at baseline predicted higher end of semester GPA
frame covers the entire “drop period.” The three-month time frame was scores. Note that TOI also was not a significant predictor of GPA when
also chosen because it has been used in prior research testing the effects entered in the absence of CSQ score and vice versa. Similarly, an ex-
of attributional style on prospective changes in depressive symptoms ploratory analysis testing the TOI × Stress interaction did not yield
(e.g., Haeffel & Hames, 2014; Sasso, Giovanetti, Schied, Burke, & statistically significant findings.
Haeffel, 2019).
At baseline (approximately 2–4 weeks after the start of the seme- 2.3. Predicting dropped courses
ster), participants were administered an informed consent form, a brief
demographics questionnaire, and measures assessing mindsets of in- We hypothesized that individuals with greater entity beliefs (i.e.,
telligence (TOI), negative attributional style (CSQ), perceived academic lower TOI scores) and greater levels of negative attributional style (i.e.,
stress, and depressive symptoms (BDI). Participants also self-reported higher CSQ achievement subscale scores) would be more likely to drop
their ACT scores at baseline, which was used as a covariate in analyses. a course than those with greater incremental beliefs of intelligence and
At the end of the semester, approximately 3 months later, the partici- less negative attributional style. Logistic regression was used to test the
pants were administered a questionnaire asking if they had dropped any hypothesis (see Table 2). Results showed that baseline GPA and CSQ
courses and were again administered the measures of academic stress achievement subscale scores were the only significant predictors of
and depressive symptoms. One hundred twenty-five of 130 participants whether or not a student dropped a course. Those with lower GPA
completed all Time 1 and Time 2 measures at both time points (96%). scores were more likely to drop a course than those with higher GPA
Completers and non-completers did not differ on any of the baseline scores. And, those with higher CSQ scores were more likely to drop a
measures. course than those with lower CSQ scores. Note that TOI also was not a
significant predictor of dropping a course when entered in the absence
of CSQ score. Similarly, an exploratory analysis testing the TOI × Stress
2. Results interaction did not yield statistically significant findings.

Means, standard deviations, and inter-correlations of study mea-

2.4. Predicting depressive symptoms
sures are summarized in Table 1. An analysis of the distribution of
scores (see Fig. 1) shows that there were a relatively large range of
We hypothesized that individuals with greater entity beliefs (i.e.,
scores on both measures.
lower TOI scores) and greater levels of negative attributional style (i.e.,
higher CSQ achievement subscale scores) would be more likely to ex-
perience increases in depressive symptoms than those with greater in-
2.1. Correlation between intelligence mindset and negative attributional
cremental beliefs of intelligence and less negative attributional style.
style
Hierarchical multiple regression was used to test the hypothesis (see
Table 2; note that change in perceived stress from pre-to-post was op-
Contrary to hypotheses, TOI scores and CSQ achievement scale
erationalized as a standardized residual from regressing T2 perceived
scores were not highly correlated (r = −0.21; shared variance = 4%)
stress scores onto T1 perceived stress scores). Results showed that
indicating that they are likely measuring distinct constructs.
baseline depression scores and the negative attributional style-stress

Table 1
Means, standard deviations, and correlations.
1 2 3 4 5 6 7 8

1 ACT score –
2 GPA 0.42 –
3 TOI −0.14 0.00 –
4 CSQ 0.16 0.05 −0.21 –
5 BDI T1 −0.02 0.05 −0.10 0.32 –
6 Stress T1 −0.14 −0.19 −0.04 0.13 0.22 –
7 BDI T2 −0.04 0.10 −0.17 0.28 0.70 0.16 –
8 Stress T2 −0.06 −0.14 −0.04 0.02 0.18 0.44 0.17 –
9 Dropped course −0.05 −0.24 −0.09 0.22 0.22 0.08 0.15 0.30 –
Mean 32.56 3.48 3.72 4.13 9.00 0.71 6.61 0.75 0.21
SD 2.00 0.35 0.98 0.81 6.71 0.46 5.87 0.43 0.41

Note. N = 130. ACT = American College Testing score; GPA = Grade Point Average; TOI = Theory of Intelligence measure; CSQ = Cognitive Style Questionnaire
Achievement Subscale; BDI T1 = Beck Depression Inventory at Time 1; Stress T1 = Perceived academic stress at Time 1; BDI T2 = Beck Depression Inventory at
Time 2; Stress T2 = Perceived academic stress at Time 2.
Higher scores on the TOI indicate greater incremental beliefs (i.e., intelligence is malleable). Higher scores on all other measures indicate greater levels of the
construct being measured. Correlations in bold are significant at the 0.05 level.

4
A.I. Alatorre, et al. Learning and Individual Differences 77 (2020) 101811

Fig. 1. Distribution of scores on the Theory of Intelligence (TOI) and the Cognitive Style Questionnaire (CSQ).

interaction were significant predictors of prospective changes in de- prospective interval (controlling for initial symptom levels). Note that
pressive symptoms. As expected, those with greater baseline depression the TOI × Stress interaction was not a significant predictor of pro-
scores also had greater levels of depression at time 2. To examine the spective changes in depressive symptoms.
interaction effect, we conducted a simple slope analysis (Cohen, Cohen,
West, & Aiken, 2003) which showed that the gradient of the simple
2.5. Post-hoc analysis
slope for those with “high” and “low” levels of perceived stress was
significantly different depending on level of negative attributional style
Prior work suggests that mindset theories of intelligence are most
(t = 17.33; p < .001). As predicted, participants with high levels of
likely to show an effect on academic outcomes in low academically
negative attributional style and an increase in perceived stress experi-
achieving samples. Thus, we conducted an exploratory analysis in
enced the greatest levels of depressive symptoms at the end of the
which we examined the interaction of TOI and student achievement

5
A.I. Alatorre, et al. Learning and Individual Differences 77 (2020) 101811

Table 2 mindset, are related to attributional style and the general factor found
Cognitive vulnerability and theory of intelligence predicting academic and by Schroder et al. (2016).
depressive symptom outcomes. In our study, TOI did not demonstrate incremental or predictive
Predictor b SE pr t p R2 change validity. Intelligence mindset was not a significant predictor of aca-
demic outcomes (end of semester GPA or dropping a class) or future
DV: semester GPA at Time 2 depressive symptoms. This finding is consistent with at least a handful
Step 1 0.34⁎
of other studies (e.g., Dixson, Roberson, & Worrell, 2017; Furnham,
ACT 0.04 0.02 0.22 1.91 0.06
GPA T1 0.55 0.11 0.52 5.27 < .001 Chamorro-Premuzic, & McDougall, 2003; Hwang, Reyes, & Eccles,
Step 2 0.01 2019; Rienzo, Heather, & Wilkinson, 2015) and two recent meta-ana-
CSQ −0.01 0.04 −0.03 −0.25 0.80 lyses showing that in experimental designs in which intelligence
TOI 0.02 0.03 0.07 0.60 0.55
mindset is directly manipulated, its effect on future academic outcomes
DV: dropped a course tends to be small. Effect sizes in these studies are particularly small for
Step 1 Wald Exp(B) students performing at a high academic level. For example, Paunesku
ACT 0.04 0.15 0.08 1.04 0.78
GPA T1 −1.79 0.73 6.06 0.17 0.01
et al. (2015) found that mindset interventions tend to be most beneficial
Step 2 for poorly-performing students. Similarly, Yeager et al. (2018) found
CSQ 0.88 0.35 6.34 2.41 0.01 that an online growth-mindset intervention was only effective for stu-
TOI −0.19 0.26 0.53 0.83 0.93 dents who were at significant risk for compromised well-being. These
DV: depressive symptoms at T2 findings are consistent with the conclusions of those of Sisk et al. (2018)
Step 1 0.51⁎ who found a weak relationship between mindset and academic
ACT −0.22 0.24 −0.08 -.91 0.37
achievement, but stated that mindset interventions may still be effec-
GPA T1 1.72 1.23 0.13 1.39 0.17
BDI T1 0.61 0.06 0.71 10.79 < .001
tive for at-risk subgroups (see also Burns & Isbell, 2007). Thus, it is
Step 2 0.02 possible that the TOI might have performed better if we had tested our
CSQ 0.63 0.52 0.11 1.21 0.23 hypotheses in a sample that was less academically high-achieving
TOI −0.55 0.28 −0.13 −1.38 0.17 (Hwang et al., 2019). However, recent work suggests that even among
Stress change 0.40 0.38 0.10 1.04 0.30
low-achieving samples, the effect of mindset may not add incrementally
Step 3 0.02⁎
CSQ × stress 1.22 0.50 0.22 2.43 0.02 to the prediction of achievement when controlling for socioeconomic
change status (SES). Specifically, Dixson et al. (2017) noted that the impact of
growth mindset on achievement may be “exaggerated” as a result of not
Note. ACT = American College Testing score; GPA T1 = Overall Grade Point controlling for SES.
Average at T1; TOI = Theory of Intelligence measure; CSQ = Cognitive Style The CSQ was a significant predictor of depressive symptoms, which
Questionnaire Achievement Subscale; BDI T1 = Beck Depression Inventory at
is consistent with prior research (see review by Haeffel et al., 2008).
Time 1; Stress Change = Perceived stress residual change score.
⁎ This was among the first studies to show that the CSQ could also predict
p < .05.
academic related outcomes. Specifically, the CSQ achievement subscale
was a significant predictor of whether or not participants dropped a
(participants who scored in the bottom half of the sample on both GPA
course (even after controlling for initial GPA and ACT scores). We
and ACT scores versus those who scored in the top half of the sample on
suspect that the effect of the CSQ might be mediated by hopelessness.
both GPA and ACT). Results showed that the effect of intelligence
Students with a negative attributional style become hopelessness about
mindset on future GPA (and dropping a course) did not change as a
their ability to improve their grade and, in turn, drop the course. It may
function of prior achievement status (t = 0.89, p = .37).
be fruitful for future research to use a three-time point study in which
possible mediators of the effect of CSQ on dropping a course (e.g.,
3. Discussion hopelessness beliefs) could be measured and tested.
These results raise the following question: why would a measure of
The purpose of the current study was to test if mindset about in- vulnerability to depression be a better predictor of academic outcomes
telligence represents a novel construct or if it should be considered a than a measure of intelligence mindset, which was specially theorized
specific instance of a more general negative attributional style as the- to be a predictor of academic outcomes? The most parsimonious ex-
orized by the cognitive theories of depression (e.g., Abramson et al., planation is that individual differences in how people think about
1978; Beck, 1976). Results showed that the mindset theory of in- academic stress is more important for predicting their achievement
telligence (as measured by the TOI) is distinct from the negative attri- (and depressive symptoms) than individual differences in how people
butional style construct featured in the hopelessness theory of depres- think about the stability of their intelligence. Indeed, the results of the
sion (as measured by the CSQ). However, the TOI did not demonstrate current study combined with recent experimental work raises concerns
predictive or incremental validity. The TOI did not prospectively pre- about the predictive validity of intelligence mindset. However, as
dict end of semester GPA, dropping a class, or changes in depressive Swann, Chang-Schneider, and McClarty (2007) warn, it may be pre-
symptoms. The CSQ achievement scale predicted dropping a class and mature to totally dismiss intelligence mindsets before considering the
prospective changes in depressive symptoms during times of perceived “match” between the TOI and the outcomes to be predicted. Under
academic stress. what conditions might we expect a fixed mindset about intelligence be
Contrary to predictions, TOI scores were weakly correlated with most predictive? According to mindset theory, an entity mindset is
CSQ scores (r = −0.2). This indicates that people's perceptions about detrimental because it is more likely than an incremental mindset to
the malleability of intelligence do not generalize to the inferences that lead people to adopt a performance rather than mastery orientation.
they generate for negative events in their life. In other words, if an This theory is typically applied to all students, regardless of contextual
individual believes that intelligence is unchangeable, it does not mean factors. However, it is possible that having an entity mindset is not
that he or she is also more likely to view the causes of other personal or always maladaptive. For example, if someone has high intelligence (or
environmental variables as unchangeable. These results indicate that perceives being of high intelligence) and is generally doing well aca-
mindset about intelligence is a distinct construct from the negative demically, then having the fixed mindset may actually lead them to
attributional style that confers risk for depression. That said, it remains persist during adversity because they are confident in their intellectual
unclear if other mindset domains (e.g., beliefs about the stability of abilities (Castles, 2012; Eccles & Wigfield, 2002). However, if one is of
personality or mental illness), which may not overlap with intelligence low intelligence (or perceives themselves as low intelligence), then

6
A.I. Alatorre, et al. Learning and Individual Differences 77 (2020) 101811

believing intelligence is fixed is likely to be detrimental during times of org/content/dam/act/unsecured/documents/CCCR_National_2016.pdf.

adversity because they are helpless to improve their intellectual abil- Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New York: International
Universities Press.
ities. This reasoning is consistent with research indicating that the Beck, A. T., Rush, A. J., Shaw, B. F., & Emery, G. (1979). Cognitive therapy of depression.
mindset theories of intelligence are most likely to show an effect on New York: Guilford Press.
academic outcomes in low academically achieving samples (perhaps Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck
Depression Inventory: Twenty-five years of evaluation. Clinical Psychology Review, 8,
those who question their intellectual abilities). That said, the post-hoc 77–100.
exploratory analysis in which we examined the interaction of TOI and Blackwell, L. S., Trzesniewski, K., & Dweck, C. S. (2007). Implicit theories of intelligence
student achievement (participants who scored in the bottom half of the predict achievement across an adolescent transition: A longitudinal study and an
intervention. Child Development, 78, 246–263. https://doi.org/10.1111/j.1467-8624.
sample on both GPA and ACT scores versus those who scored in the top 2007.00995.x.
half of the sample on both GPA and ACT) showed that the effect of Burnette, J. L., O’Boyle, E., VanEpps, E. M., Pollack, J. M., & Finkel, E. J. (2013). Mindsets
intelligence mindset on academic outcomes did not change as a func- matter: A meta-analytic review of implicit theories and self-regulation. Psychological
Bulletin, 139(3), 655–701. https://doi.org/10.1037/a0029531.
tion of prior achievement status. However, it is important to remember
Burns, K. C., & Isbell, L. M. (2007). Promoting malleability is not one size fits all: Priming
that the “low-achieving” subset of in the current sample is still higher implicit theories of intelligence as a function of self-theories. Self and Identity, 6,
achieving than most students at in United States colleges. 51–63.
This study had both strengths and limitations. It was the first to test Castles, E. E. (2012). Inventing intelligence: How American came to worship IQ. Santa Bara,
CA: PraegerABC-CLIO.
the incremental validity of the TOI measure over what was hypothe- Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/cor-
sized to be a highly similar construct. Such comparisons have the po- relation analysis for behavioral sciences (3rd ed.). Mahwah, NJ: Lawrence Erlbaum
tential to constrain the production of overlapping constructs and the- Associates, Inc.
Dixson, D. D., Roberson, C. B., & Worrell, F. C. (2017). Psychosocial keys to African
ories as well as indicate points for integration. Further, the study used a American achievement? Examining the relationship between achievement and psy-
prospective longitudinal design, which enabled us to establish temporal chosocial variables in high achieving African Americans. Journal of Advanced
precedence for the effect of mindset theory and negative attributional Academics, 28, 120–140.
Dweck, C. S. (1986). Motivational processes affecting learning. American Psychologist, 41,
style on two academic outcomes as well as changes in depressive 1040–1048.
symptoms. The study also had limitations. First, we used a sample of Dweck, C. S. (1999). Essays in social psychology. Self-theories: Their role in motivation,
highly motivated and high-achieving undergraduate students, which personality, and development. New York, NY, US: Psychology Press.
Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to motivation and
raises questions about generalizability of our findings to at-risk student personality. Psychological Review, 95, 256–273. https://doi.org/10.1037/0033-295X.
populations. Further, we only examined outcomes for a single academic 95.2.256.
semester. It is possible that the effect of an entity mindset of intelligence Eccles, J. S., & Wigfield, A. (2002). Motivational beliefs, values, and goals. Annual Review
of Psychology, 53, 109–132.
could emerge over a longer prospective interval. Additionally, the
Ellis, A. (1962). Reason and emotion in psychotherapy. New York: Lyle-Stuart.
current study used a measure of stress that consisted of a single self- Furnham, A., Chamorro-Premuzic, T., & McDougall, F. (2003). Personality, cognitive
report item. It will be useful for future research to use a measure of ability, and beliefs about intelligence as academic performance. Learning and
stress that is more comprehensive in scope. Finally, the sample tended Individual Differences, 14, 49–66.
Haeffel, G. J. (2010). Cognitive vulnerability to depressive symptoms in college students:
to have low levels of depressive symptoms, so it is unclear if results A comparison of traditional, weakest-link, and flexibility operationalizations.
would generalize to populations with clinically significant levels of Cognitive Therapy and Research, 34, 92–98.
depression. Haeffel, G. J., Gibb, B. E., Metalsky, G. I., Alloy, L. B., Abramson, L. Y., Hankin, B. L., ...
Swendsen, J. D. (2008). Measuring cognitive vulnerability to depression:
The results may have practical implications. If colleges want to Development and validation of the cognitive style questionnaire. Clinical Psychology
improve academic outcomes, then it may be preferable to target ne- Review, 28(5), 824–836.
gative attributional style rather than intelligence mindsets. Reducing Haeffel, G. J., & Hames, J. L. (2014). Cognitive vulnerability to depression can be con-
tagious. Clinical Psychological Science, 2, 75–85.
negative attributional styles is more likely to benefit students emo- Hankin, B. L., Lakdawalla, Z., Carter, I. L., Abela, J. R. Z., & Adams, P. (2007). Are
tionally and academically (e.g., Seligman, Schulman, DeRubeis, & neurotocism, cognitive vulnerabilities and self-esteem overlapping or distinct risks
Hollon, 1999) than changing their beliefs about intelligence. That said, for depression? Evidence from exploratory and confirmatory factor analyses. Journal
of Social and Clinical Psychology, 26, 29–63.
it may be difficult to overcome the popularity of the fixed mindset
Heiligenstein, E., & Guenther, G. (1996). Depression and academic impairment in college
hypothesis. The mindset hypothesis has received an incredible amount students. Journal of American College Health, 45, 59–64.
of attention from administrators, teachers, and laypeople. A Google Henderson, V. L., & Dweck, C. S. (1990). Motivation and achievement. In S. S. Feldman, &
G. R. Elliott (Eds.). At the threshold: The developing adolescent (pp. 308–329).
search for intelligence mindsets has > 100 times the number of hits
Cambridge, MA: Harvard University Press.
(45,000,000 hits) as a search for negative attributional style (400,000 Hong, Y. C., Chiu, C. S., Dweck, D. M. S., Lin, W., & Wan, W. (1999). Implicit theories,
hits). Thus, getting universities to attend to the sizeable body of re- attributions, and coping: A meaning system approach. Journal of Personality and
search on attributional style may be difficult relative to their exposure Social Psychology, 77, 588–599.
Hwang, N., Reyes, M., & Eccles, J. S. (2019). Who holds a fixed mindset and whom does it
to articles on intelligence mindsets. harm in mathematics. Youth & Society, 51, 247–267.
In conclusion, this study examined whether mindset theories are a Hysenbegasi, A., Hass, S., & Rowland, C. R. (2005). The impact of depression on the
specific instance of the vulnerability construct featured in the cognitive academic productivity of university students. The Journal of Mental Health Policy and
Economics, 8, 145–151.
theories of depression or a truly unique construct. Results showed that King (2017). A fixed mindset leads to negative affect: The relations between implicit
the mindset theory of intelligence is distinct from negative attributional theories of intelligence and subjective well-being. Zeitschrift fur Psychologie, 225,
style. However, intelligence mindset, as measured by the TOI, did not 137–145.
Meehl, P. E. (1978). Theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the
exhibit predictive or incremental validity for academic and depressive slow progress of soft psychology. Journal of Consulting and Clinical Psychology, 46,
outcomes. This finding adds to a growing literature suggesting that 806–834.
trying to change entity mindset theories of intelligence may have lim- Miu, A. S., & Yeager, D. S. (2014). Preventing symptoms of depression by teaching
adolescents that people can change: Effects of a brief incremental theory of person-
ited real world value.
ality intervention at 9-month follow-up. Clinical Psychological Science, 3(5), 726–743.
https://doi.org/10.1177/2167702614548317.
References Molden, D. C., & Dweck, C. S. (2006). Finding “meaning” in psychology: A lay theories
approach to self-regulation, social perception, and social development. The American
Psychologist, 61, 192–203. https://doi.org/10.1037/0003-066x.61.3.192.
Abramson, L. Y., Metalsky, G. I., & Alloy, L. B. (1989). Hopelessness depression: A theory- Nolen-Hoeksema, S., Girgus, J. S., & Seligman, M. E. P. (1986). Learned helplessness in
based subtype of depression. Psychological Review, 96(2), 358–372. https://doi.org/ children: A longitudinal study of depression, achievement, and explanatory style.
10.1037/0033-295X.96.2.358. Journal of Personality and Social Psychology, 51, 435–442.
Abramson, L. Y., Seligman, M. E., & Teasdale, J. D. (1978). Learned helplessness in hu- Paunesku, D., Walton, G. M., Romero, C., Smith, E. N., Yeager, D. S., & Dweck, C. S.
mans: Critique and reformulation. Journal of Abnormal Psychology, 87(1), 49–74. (2015). Mindset interventions are a scalable treatment for academic under-
https://doi.org/10.1037/0021-843X.87.1.49. achievement. Psychological Science, 26, 785–793. https://doi.org/10.1177/
ACT, Inc (2016). The condition of college and career readiness 2016. http://www.act. 0956797615571017.

7
A.I. Alatorre, et al. Learning and Individual Differences 77 (2020) 101811

Peterson, C., & Barrett, L. C. (1987). Explanatory style and academic performance among achievement? Two meta-analyses. Psychological Science, 29, 549–571. https://doi.
university freshmen. Journal of Personality and Social Psychology, 53, 603–607. org/10.1177/0956797617739704.
Popper, K. R. (1959). The logic of scientific discovery. London: Hutchinson of London. Swann, W. B., Chang-Schneider, C., & McClarty, K. L. (2007). Do people’s self-views
Rienzo, C., Heather, R., & Wilkinson, D. (2015). Changing mindsets: Evaluation report and matter? Self-concept and self-esteem in everyday life. American Psychologist, 62,
executive summary. London, UK: Education Endowment Foundation. https://v1. 84–89.
educationendowmentfoundation.org.uk/uploads/pdf/Changing_Mindsets.pdf. Tsujimoto, K. C., Frijters, J. C., Boada, R., Gottwald, S., Hill, D., Jacobson, L. A., & Bosson-
Rojstaczer, Stuar, S., & Healy, C. (2010). Grading in American colleges and universities. Heenan, J. (2018). Achievement attributions are associated with specific rather than
Teachers College Record, 1–6. general learning delays. Learning and Individual Differences, 64, 8–21. https://doi.org/
Sasso, M. P., Giovanetti, A. K., Schied, A. L., Burke, H. H., & Haeffel, G. J. (2019). #Sad: 10.1016/j.lindif.2018.04.002.
Twitter content predicts changes in cognitive vulnerability and depressive symptoms. Weiner, B. (1985). An attributional theory of achievement motivation and emotion.
Cognitive Therapy and Research, 43, 657–665. Psychological Review, 92, 548–573.
Schleider, J. L., Abel, M. R., & Weisz, J. R. (2015). Implicit theories and youth mental Wilson, T. D., & Linville, P. W. (1982). Improving the academic performance of college
health problems: A random-effects meta-analysis. Clinical Psychology Review, 35, 1–9. freshmen: Attribution therapy revisited. Journal of Personality and Social Psychology,
https://doi.org/10.1016/j.cpr.2014.11.001. 42(2), 367–376.
Schroder, H. S., Moran, T. P., Donnellan, M. B., & Moser, J. S. (2016). Mindset induction Yeager, D. S., Hanselman, P., Paunesku, D., Hulleman, C., Dweck, C., Muller, C., ...
effects on cognitive control: A neurobehavioral investigation. Biological Psychology, Duckworth, A. (2018, March 1). Where and for whom can a brief, scalable mindset
103, 27–37. https://doi.org/10.1016/j.biopsycho.2014.08.004. intervention improve adolescents’ educational trajectories? https://doi.org/10.31234/
Seligman, M. E. P., Schulman, B. S., DeRubeis, R. J., & Hollon, S. D. (1999). The pre- osf.io/md2qahttps://psyarxiv.com/md2qa/.
vention of depression and anxiety. Prevention & Treatment, 2, 1–24. Zhang, Z., & Yuan, K.-H. (2015). WebPower: Statistical power analysis online. Retrieved
Sisk, V. F., Burgoyne, A. P., Sun, J., Butler, J. L., & Macnamara, B. N. (2018). To what from http://webpower.psychstat.org.
extent and under which circumstances are growth mind-sets important to academic